Production agent architecture

Our agent is too unreliable to operate

The harness—not the model—owns reliable production behavior.

Non-deterministic answers are only one source of agent failure. Lost work, duplicated effects, ambiguous state, unreconstructable incidents, and unsafe retries are failures of the runtime around the model.

Reliability comes from explicit control flow, contracts, checkpoints, idempotent effects, telemetry, and recovery. Prompting the model to behave consistently cannot supply those operating guarantees.

See the Production Agent Architecture pillar →

Decisions that make behavior operable

  1. Control flow: Keep lifecycle, budgets, approvals, and effects deterministic; use model judgment only inside bounded steps.
  2. Contracts: Validate tool requests, results, state transitions, timeouts, and failure states explicitly.
  3. Effects: Give consequential calls idempotency keys and explicit commit boundaries so retries do not duplicate work.
  4. Recovery: Record checkpoints and cancellation events so interrupted runs can be replayed from known state.
  5. Evidence: Preserve decisions, attempts, outcomes, and request identity well enough to reconstruct an incident.

Read the reliability path

Proof and limits


Review the contracts around your agent

A focused review traces one consequential trajectory through control flow, tools, state, authority, verification, effects, and recovery to identify the missing contracts and sequence remediation.